Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
1
Pragmatic aspects of implicit comparison: an economy-based approach
Osamu Sawada
University of Chicago
Journal of Pragmatics 41: 1079-1103 (2009), Pre-publication version.
E-mail Address
Abstract
There are two ways to express comparison: implicit comparison (e.g. Compared to Tom, Jim
is tall.) and explicit comparison (e.g. Jim is taller than Tom.) (Sapir, 1944; Kennedy, in
press). Although implicit comparison and explicit comparison can both be used to express
comparison, they have different pragmatic properties. In these examples, the former, but
not the latter, implies (a) Tom is not tall and (b) Jim is not definitely tall (possibly
borderline). This paper investigates the pragmatic aspects of implicit comparison
cross-linguistically (i.e. English and Japanese), considering (1) how it is pragmatically
different from explicit comparison, (2) the status of the two implicatures, and (3) how they
arise.
There are two approaches to explain the implicatures in implicit comparison: a
symmetrical (or economy-based) approach, and an asymmetrical (or dependency) approach.
In the symmetrical approach, the two implicatures are viewed as deriving from a single
processing principle of economy. Under the asymmetrical approach, the implicature in (b)
depends on the implicature in (a). I argue that the symmetrical approach is preferable.
Since the economy principle is neither stored in a particular lexicon, nor is it a
co-operative principle, the implicatures have to be regarded as belonging to a third type of
implicature: „computational implicature.‟
Key words: implicit comparison, explicit comparison, context-dependency, implicature,
economy of standard, truth-value, scale structure
2
1 Introduction
Observe the following sentences:
(1) a. Compared to Jim, Tom is tall.
b. Tom is taller than Jim.
Building on the discussion of gradability in Sapir (1944), Kennedy (in press) proposes that
there are two modes of comparison, implicit comparison (=1a) and explicit comparison (=1b)
and argues that the distinction between them can be attributed to the parameters of
comparison. Roughly speaking, in English an explicit comparison is one in which a
comparative morphology (-er/more) is used, while an implicit comparison is one in which an
unmodified positive form of a gradable predicate is used. Kennedy (in press) argues that
although implicit comparison and explicit comparison can both be used to express
comparison, they differ in their semantics. Explicit comparison establishes the ordering
between the two objects, x and y, using a degree morphology, not a contextually determined
standard. Implicit comparison, on the other hand, compares the two objects, x and y, by
manipulating the context in such a way that the positive form of a gradable predicate is true
of x and false of y.
It is important to note that these two modes are different in pragmatics as well as in
semantics. (2a), but not (2b), implies (3a) and (3b):
(2) a. Compared to Tom, Jim is tall. (implicit comparison)
b. Jim is taller than Tom. (explicit comparison)
(3) a. Tom is not tall.
b. Jim is not definitely tall. (possibly borderline)
The purpose of this paper is to examine the pragmatic aspects of implicit comparison as in
(2a) and consider (a) how implicit comparison is pragmatically different from explicit
comparison, (b) the status of the implicatures, and (c) where the implicatures of implicit
comparison come from.
Let us observe the implicatures of implicit comparison. (3a) shows that unlike explicit
comparison, implicit comparison implies that the „standard of comparison‟, in this case Tom,
is low on the scale associated with the gradable predicate in the main clause. I will call this
„the implicature (inference) from the standard.‟ The existence of this implicature is clarified
by the contrast in (4):
3
(4) a. Compared to {a gymnast/?? a basketball player}, Tom is tall.
b. Compared to {a gymnast/ a basketball player}, Tom is taller.1
c. Tom is taller than {a gymnast/a basketball player}.
In (4a), but not in (4b) and (4c), the sentence with the NP a basketball player is odd because
the common knowledge that basketball players are tall conflicts with the low scale
implicture for the standard of comparison.
The second implicature in (3b), on the other hand, shows that the (unmodified) positive
form is false of the subject Jim. I will call this implicature „the implicature from the main
clause.‟2 Notice, however, that the truth value of the main clause does not always have to be
clearly false. It can be a borderline case:
(5) (Context: Mary is asked whether Jim is tall, but she cannot decide whether he is tall
or not.)
Mary: Well, compared to Tom, Jim is tall.
(a) → Jim is not tall.
(b) → I am not sure whether Jim is tall.
In this context, the speaker seems to imply (5b) rather than (5a). The implicature in (5b) is
1 Some native speakers might think that (4b) is not perfectly natural. This is presumably
due to the availability of and preference for (4c) (Chris Kennedy p.c.). 2 Notice that the following sentences are different from the compared to construction that is
the focus of this paper:
(i) Britain will have 46,180 tonnes in 1990, compared to 55,600 this year. (BNC)
(ii) More than half of working men with degree qualifications work in professional and
managerial jobs compared to 35 per cent of women. (BNC)
I am assuming that the phrase compared to in (i) and (ii) has the function of contrastiveness.
There are no implicatures like (3a) and (3b) in (i) and (ii). I will call this type of compared to
the contrastive compared to construction. The compared to construction that we are going to
focus on in our paper and the contrastive compared to construction are different in syntax,
as well. Although compared to can be fronted as in (iiib), it is not natural to front compared to in the contrastive compared to construction, as shown in (ivb) (Chris Kennedy p.c.):
(iii) a. Jim is tall compared to Tom.
b. Compared to Tom, Jim is tall.
(iv) a. Britain will have 46,180 tonnes in 1990, compared to 55,600 this year. (BNC)
b. ?? Compared to 55,600 this year, Britain will have 46,180 tonnes in 1990.
These differences suggest that the compared to construction and the contrastive compared to construction are, in fact, different constructions.
4
epistemically weaker than that in (5a). The example in (5) suggests that the negative
implicature is not uniform.
The above discussion strongly suggests that it is necessary to distinguish between
implicit and explicit comparisons in pragmatics, as well as in semantics. However, the
distinction between implicit and explicit comparisons proves not to be straightforward, if we
take into consideration languages that do not have a specialized comparative morphology
like the English -er/more.3 For example, Japanese is a language which does not use
comparative morphology in the environment „x is A-er than y, as shown in (6):‟4
(6) Taro-wa Ziro-yori se-ga takai.
Taro TOP Ziro-than back- NOM high
„Taro is taller than Ziro.‟
3 Ultan (1972) states that 32 of 108 languages surveyed by him do not have overt
comparative morphology.
4 The adverb motto is often translated as „more‟, but it cannot be regarded as a pure
comparative morpheme:
(i) Taro-wa Ziro-yori motto kasikoi.
Taro-TOP Ziro-than MOTTO smart
„Taro is even smarter than Ziro.‟
In (i) the speaker thinks that both Taro and Ziro are construed as smart (See also Beck et al.,
2004: 327). In this sense, motto is similar to English even.
Note that in Modern Japanese, yori can behave like English –er/more in special
environments:
(ii) Yori ooku-no nihon-jin-ga Denver-yori New York-ni sun-deiru.
More many-GEN Japanese people-NOM Denver-than New York-LOC live-STATIVE
„More Japanese people are living in New York than in Denver.‟
(iii) Taro-wa yori anzenna tokoro-ni hikkoshi-ta.
Taro-TOP more safe place-DAT move-PAST
„Taro moved to a safer place.‟
As Martin (1975) and many Japanese dictionaries point out, Modern Japanese has
developed the new comparative morpheme yori, corresponding to English „more‟, under the
influence of translations from European languages. Notice however that the comparative
morpheme yori is not used freely. As Sawada (in press) argues, the comparative morpheme
yori is used only if a given sentence cannot otherwise express comparison. For example, the
comparative morpheme yori in (ii) is necessary because if we delete it, (ii) is interpreted as a
sentence with the expression „rather than‟, i.e. „Many Japanese people are living in New
York rather than in Denver.‟ Furthermore, the comparative morpheme is necessary for
comparison in (iii) as well. This is because if it is deleted, (iii) must be interpreted as a
simple sentence with a bare adjective, i.e. „Taro moved to a safe place.‟
What is crucial here is that the comparative morpheme yori in (6) is not necessary
because we can express comparison without it. The corpus data also show that the
comparative morpheme yori is rarely used in environments where we can express
comparison without it. In this paper, we will only focus on environments in which the
comparative morphology does not normally appear, i.e. „x is A-er than y.‟
5
This might lead one to think that Japanese does not have a mode of explicit comparison,
since it has no specialized morphology that expresses an arbitrary ordering relationship, an
idea considered by Beck et al. (2004). However, as Kennedy (in press) argues, the Japanese
comparative yori does allow explicit comparison, and is semantically different from the
implicit comparatives kurabe-tara „compare-conditional‟ or kurabe-ru-to
„compare-present-conditional.‟ It is important to notice that in Japanese, implicit and
explicit comparisons are differentiated in the comparative clause/phrase. That is to say,
implicit comparison is expressed by the conditional morpheme -tara or -ru-to with the verb
stem kurabe „compare,‟ as in kurabe-tara „compare-conditional‟ or kurabe-ru-to
„compare-present-conditional.‟5 As Sawada (2005, 2007) points out, these expressions are
pragmatically different from the yori- comparative or kurabe(-te) „compare-TE,‟ although
the main clauses of implicit and explicit comparisons in Japanese have exactly the same
forms, as in (8) and (9). (7) is a simple sentence with an adjective:
(7) Taro-wa se- ga takai.
Taro-TOP height-NOM tall
„Taro is tall.‟
(8) a. Explicit comparison
Taro-wa Ziro-yori se- ga takai.
Taro TOP Ziro-than back- NOM high
„Taro is taller than Ziro.‟
b. Explicit comparison
Ziro-ni kurabe(-te) Taro-wa se-ga takai.6
Ziro-DAT compare-TE Taro-TOP height-NOM tall
„Taro is taller than Ziro.‟
(9) a. Implicit comparison
Ziro-ni kurabe-tara Taro-wa se-ga takai.
Ziro-DAT compare-COND Taro-TOP height-NOM tall
„Compared to Ziro, Taro is tall.‟
5 Kurabe-reba „compare-conditional‟ also behaves as implicit comparison. Thanks to Yusuke
Kubota for bringing this to my attention. 6 Notice that here te can be omitted. I will gloss the clause linker –te as TE. It seems to me
that -te in kurabe(-te) is not to be interpreted as „and.‟ See also Yuasa and Sadock (2002) for
a detailed discussion of this Japanese clause linker.
6
b. Implicit comparison
Ziro-ni kurabe-ru-to Taro-wa se- ga takai.
Ziro-DAT compare-PRES-COND Taro-TOP height-NOM tall
„Compared to Ziro, Taro is tall.‟
Although the examples in (8) and (9) both have the same main clause, only the ones in (9)
imply (10a) and (10b):
(10) a. Ziro is short.
b. Taro is not definitely tall. (possibly a borderline)
These examples suggest that the distinction between explicit and implicit comparisons does
not come from the superficial form of an adjective, nor from the presence or absence of the
verb compare, because (a) there is no comparative morpheme in (8) and (9), and (b)
Japanese allows us to express implicit and explicit comparisons using the verb stem kurabe-
„compare.‟
Given these empirical facts, how can we explain the pragmatic aspects of implicit
comparison theoretically? Specifically, what is the status of the two implicatures in implicit
comparison? Where do they come from? Why do they have „low‟ instead of „high‟ scalar
inference? Why does a speaker use an implicit comparison (e.g. Compared to Tom, Jim is
tall.) rather than a simple form of a gradable predicate (e.g. Jim is tall. /Jim is not tall.) in a
given context?
This paper tries to answer these questions by comparing two different approaches. One
approach is what I will call the „symmetrical‟ (or economy-based) approach and the other
approach is what I will call „asymmetrical‟ (or dependency) approach. In the symmetrical
approach, the two implicatures in implicit comparison are viewed as deriving from a single
processing principle, the economy of standard. In this view, both of the implicatures have
the same status because they are derived simultaneously.
In the asymmetrical approach, on the other hand, the two implicatures in implicit
comparison are considered to have a different status from each other. This approach is
asymmetrical because in it, the derivation of the implicature from the main clause is
considered to „depend on‟ the existence of the implicature from the standard. Although both
approaches work, I will argue that the symmetrical approach has more advantages than the
asymmetrical approach.
This paper proceeds as follows. Section 2 introduces the analyses of explicit comparison
and implicit comparison based on Kennedy (in press). Section 3 adduces the empirical
7
differences between the two modes of comparison in English and Japanese. Section 4
discusses the two implicatures in implicit comparison and provides a preliminary analysis.
Section 5 introduces the symmetrical approach to the implicatures of implicit comparison,
and Section 6 introduces the alternative asymmetrical approach to them. Section 7
compares the two approaches and argues that the symmetrical approach has more
advantages than the asymmetrical approach. Section 8 is the conclusion.
2. Semantics of explicit comparison and implicit comparison
Before embarking on a pragmatic account of implicit comparison, let us briefly look at the
semantics of explicit and implicit comparison. As we saw in Section 1, explicit comparison
establishes an ordering between the two objects x and y with respect to gradable property g
using a degree morphology. The conventional meaning of such explicit comparison is that
the degree to which x is g exceeds the degree to which y is g. There are various ways to
represent the semantics of (11a), but here I will represent it as (11b):
(11) a. x is A-er than y.
b. The degree to which x is A > the degree to which y is A
(11b) says that the degree to which x is tall exceeds the degree to which y is tall. The
important point is that the semantics of (11a) establishes the ordering of x and y with
respect to the scale of A, but does not tell us whether x and y are actually A or not.
Next, let us consider the semantics of implicit comparison. Kennedy (2007a, in press)
argues that the function of compared to is to ensure that the denotation of the predicate is
calculated with respect to a „context‟ that includes only the two objects being compared. The
semantics of (12a) can be represented in (12b):
(12) a. Compared to y, x is A
b. [[ A compared to y ]] is true of x in a context c iff [[ A ]] is true of x in any context c‟
just like c except that the domain includes just x and y.7
7 Note that the main clause of a sentence containing compared to does not always have to
contain an adjective. The predicate can be a noun:
(i) Compared with my house, yours is a castle.
(ii) Compared with my daughter, yours seems a veritable angel.
I thank one of the reviewers for noting that these are cases of simile or metaphor. For
example, the speaker in (i) knows that the addressee‟s house is not a castle, but he/she is
8
Note that the semantics in (12) entails that x is more A than y. This is because there is a
general requirement that the domain of the predicate must always be partitioned into two
non-empty sets (a „positive‟ extension and a „negative‟ extension) (Klein, 1980; Kennedy,
2007a, in press). Thus, it follows that in a context that includes just x and y, if it is true that
x is A, it must be false that y is A, as shown in (13):
(13) Compared to y, x is A
[[ A(x)]] = 1 positive extension
[[ A(y)]] = 0 negative extension
This is equivalent to saying that the degree to which x is A exceeds the standard of
comparison in the compared to-context, but the degree to which y is A does not. It follows
from transitivity that the degree to which x is A exceeds the degree to which y is A, so (12a)
entails (11a). These are the core ideas of the semantics of implicit and explicit comparison.
The following questions immediately arise. How can we be sure that implicit comparison
and explicit comparison are semantically different? Is it really the case that Japanese has
both implicit and explicit comparison despite the fact that it does not use the overt
comparative morpheme? In order to answer these questions, we need to consider the
diagnostics for the two modes of comparison, which will help us to decide whether a
particular comparative is explicit or implicit.
3. Diagnostics for implicit vs. explicit comparison
This section adduces several diagnostics to show that explicit comparison and implicit
comparison are semantically different. We will first look at English data and then apply
these diagnostics to Japanese. These diagnostics play an especially important role for
languages like Japanese, in which there is no overt comparative morpheme.
3.1 CRISP JUDGMENT (English)
The first diagnostic is concerned with Crisp Judgment. Kennedy (in press) argues that
explicit comparison, not implicit comparison, is felicitous in contexts requiring Crisp
Judgment. These are contexts in which the difference between the two objects with respect
to the property measured by the adjective is potentially vanishingly small:
saying that it looks like one compared to his/her house. Although the truth value of yours is a castle is always false, the truth value of „what is meant‟ by a metaphor or simile can be
construed as true compared to my house.
9
(14) Context: There are two novels, a 100 page novel and a 99 page novel.
a. ?? This novel is long compared to that one.
b. This novel is longer than that one.
(15) Context: There are two novels, a 100 page novel and a 50 page novel.
a. This novel is long compared to that one.
b. This novel is longer than that one.
In the context of (14), explicit comparison is natural, but implicit comparison is odd in this
context. This is because if a 100 page novel is judged as long, a 99 page novel is also
construed as long. This novel and that novel cannot be partitioned into two non-empty sets:
negative extension and positive extension. In (15), on the other hand, both explicit and
implicit comparisons are natural. (15a) is natural because there is a significant difference
between a 100 page novel and a 50 page novel with respect to the property measured by the
adjective long; therefore, these novels can be partitioned into two different domains.
3.2 Absolute gradable adjectives (English)
The second diagnostic is concerned with absolute gradable adjectives. Gradable adjectives
can be classified into two types, absolute and relative. Absolute gradable adjectives do not
have a context-dependent standard of comparison, unlike „relative‟ gradable adjectives
(Rotstein and Winter, 2004; Kennedy and McNally, 2005; Kennedy, 2007a, in press).
Consider a sentence with the adjective bent: This rod is bent. This sentence does not mean
that the degree to which the rod is bent surpasses some contextually determined standard
of comparison, but rather simply means that the rod has a nonzero level of bentness. The
standard of comparison is fixed at the minimum degree on the scale.
Notice that it is difficult to use an absolute gradable adjective in implicit comparison in
as shown in (16b) (Kennedy, in press):
(16) a. B is more bent than A
b. ?? Compared to A, B is bent.
Figure 1
Rod A: Rod B:
10
As Figure 1 shows, (16b) is odd because both of the rods are bent and we cannot partition
them into different domains.
3.3 Differential measurements (English)
The third diagnostic is concerned with measure phrase. Kennedy (in press) argues that a
measure phrase can occur in explicit comparison but not in implicit comparison:
(17) a. Kim is 10 cm taller than Lee.
b. ?? Compared to Lee, Kim is 10 cm tall.
(17a), but not (17b), is well formed because the measure phrase (10 cm) represents the
difference between the objects compared on the same scale. (17b) is odd because we cannot
partition Kim and Lee into two non-empty sets with respect to the property of the adjective
tall.
3.4 Discourse Structures (Question and answer structures) (English)
I would like to add one more diagnostic which is concerned with discourse structure. The
distinction between implicit and explicit comparisons is reflected in the discourse structure,
as well, as shown in (18) and (19):
(18) A: Which is more expensive, this car or that car?
B‟: This car is more expensive than that car.
B”: ?? Compared to that car, this car is expensive.
(19) A: Is this car expensive?
B‟: ?? This car is more expensive than that car.
B”: Yes, compared to that car, this car is expensive.
In (18) the comparative form is natural, but the positive form is odd. In (19), on the other
hand, the comparative form is odd, but the positive form is natural.8 Empirically speaking,
implicit comparison is natural in a reply to a yes-no question but it is not for a reply to a
wh-question. There is a rule of discourse cohesion that states that, if a question is based on a
positive form, the reply must also be based on a positive form, while if a question is based on
a comparative form, the (full) answer should be based on a comparative form.
The above four diagnostics suggest that the primary function of implicit comparison is
not to simply compare the two objects on the same scale, but to judge the truth-value of the
8 The contrast may be not so strong, compared to other diagnostics.
11
proposition in the main clause by introducing a new standard. We will discuss this function
in greater detail in section 4.
3.5 Crisp Judgment (Japanese)
Now let us consider the case of Japanese comparatives. Kennedy (in press) argues that yori
behaves like explicit comparison. Japanese also can express comparison using kurabe-te,
kurabe-tara, and kurabe-ru-to. Let us compare and contrast these four kinds of
constructions using the diagnostics. As the following examples show, yori and kurabe(-te)
are felicitous in contexts requiring crisp judgment but kurabe-tara and kurabe-ru-to are not
in the context:
(20) (Context: There are two papers. One is 16 pages long and the other is 17 pages
long.)
a. Kono peepaa-wa ano peepaa-yori nagai.
This paper-TOP that paper-than long
„This paper is longer than that paper.‟
b. Ano peepaa-ni kurabe(-te) kono peepaa-wa nagai.
That paper-DAT compare-TE (linker) this paper-TOP long
„This paper is longer than that paper.‟
c. ?? Ano peepaa-ni kurabe-tara kono peepaa-wa nagai.
That paper-DAT compare-COND this paper-TOP long
„Compared to that paper, this paper is long.‟
d. ?? Ano peepaa-ni kurabe-ru-to kono peepaa-wa nagai.
That paper-DAT compare-PRES-COND this paper-TOP long
„Compared to that paper, this paper is long.‟
Kurabe-tara and kurabe-ru-to behave like the English expression compared to. (20c) and
(20d) are odd because this paper and that paper cannot be partitioned into two non-empty
sets since the difference between 100 pages and 98 pages is very small.
3.6 Differential measurement (Japanese)
The second diagnostic is concerned with differential measurement. As the following
examples show, yori and kurabe-te, but not kurabe-tara and kurabe-ru-to, allow differential
measurement phrases:
12
(21) a. Taro-wa Ziro-yori san-senti se-ga takai.
Taro-TOP Ziro-than three-centimeter height-NOM tall
„Taro is 3 cm taller than Ziro.‟
b. Ziro-ni kurabe(-te) Taro-wa san-senti se-ga takai.
Ziro-DAT compare-TE Taro-TOP three-centimeter height- NOM tall
„Taro is 3 cm taller than Ziro.‟
c. ?? Ziro-ni kurabe-tara Taro-wa san-senti se-ga takai.
Ziro-DAT compare-COND Taro-TOP three centimeter height-NOM tall
„Taro is 3 cm tall compared to Ziro.‟
d. ?? Ziro-ni kurabe-ru-to Taro-wa san-senti se-ga takai.
Ziro-DAT compare-PRES-COND Taro-TOP three-cm height-NOM tall
„Taro is 3 cm tall compared to Ziro.‟
It is important to notice that the status of the oddness in (21c) and (21d) is similar to that of
a simple sentence with the positive form. As the following sentence shows, measure phrases
cannot directly combine with the positive form of an adjective (e.g. Snyder et al., 1995;
Schwarzschild, 2005; Kennedy, in press):
(22)* Taro-wa hyaku nanaju-senti se-ga takai.
Taro-TOP hundred seventy-centimeter height- NOM tall
„Taro is 170 cm tall.‟ (Though this sentence is okay if Taro is 170 cm taller than a
contextual standard of comparison (e.g. his month-old baby.))
Therefore, the oddness of (21c) and (21d) is to be expected, since implicit comparison judges
the truth-value of the proposition in the main clause by introducing a new standard.
3.7 The noun hoo „direction‟ (Japanese)
If the noun hoo, which literally means „direction‟, is attached to the predicate form, the
sentence means that the proposition is „more or less‟ true:
(23) Kotosi- wa atatakai-hoo -da.
This year-TOP warm-direction-PRED
„It is kind of warm this year.‟
(23) means that kotosi „this year‟ belongs to the warm zone rather than to the cold zone, but
it is at the edges of that zone. The noun hoo is attached to the predicate because the speaker
13
knows that it is not particularly warm this year.9
Notice that hoo can only appear in kurabe-tara and kurabe-ru-to:
(24) a. ?? Kotosi-wa kyonen -yori atatakai-hoo-da.
This year-TOP last year-than warm-direction-PRED
„lit. This year is kind of warmer than last year.‟
b. ?? Kyonen-ni kurabe(-te) kotosi-wa atatakai-hoo-da.
Last year-DAT compare-TE this year-TOP warm-direction-PRED
„lit. This year is kind of warmer than last year.‟
c. Kyonen-ni kurabe-tara kotosi-wa atatakai- hoo-da.
Last year- DAT compare-COND this year-TOP warm-direction-PRED
„Compared to last year, this year is kind of warm.‟
d. Kyonen-ni kurabe-ru-to kotosi-wa atatakai-hoo -da.
Last year-DAT compare-PRES-COND this year-TOP warm-direction-PRED
„Compared to last year, this year is kind of warm.‟
Hoo cannot appear in yori and kurabe(-te) comparatives, because these comparatives do not
locate an object in a zone by positing a contextually determined standard. Here, we can say
that hoo has a function of hedge (Lakoff, 1972).
3.8 Absolute gradable adjectives (Japanese)
The third diagnostic is concerned with absolute gradable adjectives such as maga-tteiru
„bent‟. These can occur in yori and kurabe-te comparatives but not in kurabe-tara and
kurebe-ru-to:10
(25) a. Kono sao-wa ano sao-yori maga-tteiru.
This rod-TOP that rod-than bent-STATIVE
„This rod is more bent than that rod.‟
9 Therefore, the sentence with -hoo cannot co-occur with an intensifier:
(i) * Kotosi- wa totemo atatakai -hoo -da.
This year-TOP very warm-direction -PRED
„It is very kind of warm this year.‟
10 Since absolute gradable adjectives do not allow one to posit a contextually determined
standard, implicit comparisons with absolute gradable adjectives never induce
implicatures.
14
b. Ano sao-ni kurabe(-te) kono sao-wa maga-tteiru.
That rod-DAT compare-TE this rod-TOP bent-STATIVE
„This rod is more bent than that rod.‟
c. ?? Ano sao -ni kurabe-tara kono kasa-wa maga-tteiru.
That umbrella-DAT compare-COND this rod-Top bent-STATIVE
„Compared to that rod, this rod is bent.‟
d. ?? Ano sao-ni kurabe-ru-to kono sao-wa maga-tteiru.
That rod-DAT compare-PRES-COND this rod-TOP bent-STATIVE
„Compared to that rod, this rod is bent.‟
In the context indicated in Figure 2, (25c) and (25d) are odd because kono sao „this rod‟ is
already bent. It is not necessary to compare it with ano sao „that rod‟ in order to say that
kono sao „this rod‟ is bent.
Figure 2
Kono sao „this rod‟: Ano sao „that rod‟:
3.9 Discourse structures (Question-answer structures) (Japanese)
As the examples in (26) show, only an explicit comparison (26B1) is natural for a reply to a
wh-question:
(26) A: Taro-to Hanako-de-wa dotira-ga se- ga takai-desu-
Taro-and Hanako-PRED-TOP which-Nom back-Nom tall-PRED.POLITE-
ka ?
Q
„Who is taller Taro or Hanako?
B1: Taro-no hoo-ga (Hanako-yori) se-ga takai-desu.
Taro-Gen direction-Nom Hanako-than height-Nom tall-PRED.POLITE
„Taro is the taller of the two‟
B2: ??Hanako-ni kurabe-tara Taro-wa se-ga takai-desu.
Hanako-DAT compare-COND Taro-Top height-Nom tall-PRED.POLITE
„Compared to Hanako, Tom is tall.‟
15
B3: ??Hanako-ni kurabe-ru-to Taro-wa se-ga takai-
Hanako-DAT compare-PRES-COND Taro-TOP height-NOM tall-
desu.
PRED.POLITE
„Compared to Hanako, Tom is tall.‟
Note that (26B1) uses the special phrase no hoo. If no hoo is used with the yori-comparatives,
the sentence conventionally implies the speaker is selecting one element rather than
another. That is to say, (26B1) presupposes that there are only two choices, A and B, in the
given context.11
However, the judgment is reversed if we posit a polar question. That is, only (27B3) and
(27B4) are natural for a reply to a polar question about Taro‟s height:
(27) A: Taro- wa se-ga takai-desu-ka?
Taro- TOP back-NOM tall-PRED.POLITE-Q
„Is Taro tall?‟
B1: ??Soo desune. Taro-wa Hanako-yori se-ga takai-desu.
Let‟s see Taro-TOP Hanako-than height-NOM tall-PRED.POLITE
„Let‟s see. Taro is the taller of the two‟ (alternative comparison)
B2: ??Soo desunee. Hanako-ni kurabe(-te) Taro-wa se- ga takai-
Let‟s see. Hanako-DAT compare-TE Taro-TOP height-Nom tall-
desu.
PRED.POLITE
„Let‟s see. Taro is taller than Hanako.‟
11 Therefore, even though the following sentences are explicit comparisons, it is not natural
to use them in reply to (26A):
(i) (As a reply to (26A))
a. # Taro-wa Hanako -yori se-ga takai. (explicit comparison)
Taro-TOP Hanako-than back-NOM high
„Taro is taller than Hanako.‟
b. # Taro-wa Hanako-ni kurabe(-te) se -ga takai. (explicit comparison)
Taro-TOP Hanako-to compare-TE height-NOM tall
„Taro is taller than Hanako.‟
This is because question (26A) considers the relation between Taro and Hanako, but (ia) and
(ib) imply that Hanako is a new information.
16
B3: Soo desunee. Hanako-ni kurabe-tara Taro-wa se-ga takai-
Let‟s see Hanako-DAT compare-COND Taro-TOP height-Nom tall-
desu.
PRED.POLITE
„Let‟s see. Compared to Hanako, Taro is tall.‟
B4: Soo desunee. Hanako-ni kurabe-ru-to Taro-wa se-ga
Let‟s see. Hanako-DAT compare-PRES-COND Taro-TOP height-Nom
takai-desu.
tall-PRED.POLITE
„Let‟s see. Compared to Hanako, Taro is tall.‟
Based on these examples, we can safely conclude that kurabe-tara and kurabe-ru-to takes
the truth-value of the main proposition into account.
3.10 Summary of sections 3
The diagnostics adduced in section 3 support the argument that there are two modes of
comparison, explicit comparison and implicit comparison and they are semantically
different. We can summarize this section as follows:
(28) (Summary: descriptive)
a. In English, explicit comparison is one in which a comparative morphology
(-er/more) is used and implicit comparison is one in which an unmodified positive
form of a gradable predicate is used.
b. In Japanese, comparative morphology does not occur in a predicative use of a
gradable adjective, and yet there is a distinction between explicit comparison
and implicit comparison. The yori and kurabe(-te) constructions mark explicit
comparisons, while the kurabe-tara and kurabe-ru-to forms mark implicit
comparisons. It is likely that the conditional morphology contributes to the
semantics/pragmatics of implicit comparison in Japanese.12
d. Implicit comparison takes into account the truth-value of the proposition in the
main clause based on a linguistically expressed standard unlike explicit
12 According to one analysis of the semantics of yori and kurabe(-te), Japanese has an
invisible degree morpheme like the English –er/more that combines with a positive form of
an adjective (e.g. Beck et al., 2004). Another possibility is that there is no such thing as a
null comparative morpheme in Japanese (Kennedy, 2007b). In this view, what distinguishes
implicit and explicit comparisons in Japanese is not the semantics of the main clause, but
the semantics of the comparative clause/phrase.
17
comparison. In the latter, what matters is the ordering between a target and a
standard of comparison on a given scale.
The following figure shows the parametric variation between implicit and explicit
comparisons in English and Japanese:
Figure 3: Implicit and explicit comparisons in English and Japanese
More/-er --- explicit comparison
English
Compared to, with respect to, etc. --- implicit comparison
Yori, kurabe(-te) --- explicit comparison
Japanese
Kurabe-tara, kurabe-ru-to --- implicit comparison
One of the reviewers pointed out that the distinction between explicit comparison and
implicit comparison can be captured by the notion of the „tightness‟ of interclausal linkage.
Explicit comparison is tighter than implicit comparison. According to the reviewer,
“tightness of interclausal linkage is, in principle, positively correlated with „symbolicity‟
(decontextualization) and negatively correlated with indexicality (contextualizedness).” The
notion of tightness seems to play an important role in functional linguistics (Silverstein,
1976; Lehmann, 1988).
However, this hypothesis does not naturally explain the Japanese comparatives. In
Japanese implicit comparison and explicit comparison can both be expressed using the verb
kurabe „to compare.‟ Note that ni kurebe-te „DATE-compare-TE‟ and ni kurabe-tara
„DAT-compare-COND‟ are morphosyntactically similar in that te and tara are subordinators.
What they differ in is their semantics. That is, ni-kurabe-te behaves semantically like yori
but ni-kurabe-tara does not. 13 This indicates that the pragmatic distinction between
implicit and explicit comparison is based on a core difference in semantics.
4 Pragmatics of implicit comparison: preliminary discussion
13 Muraki (1991) and Francis and Yuasa (2008) argue that semantically, ni kurbe-te can be
analyzed as a deverbal postposition that is synonymous with yori, but that
morphosyntactically, kurabe must still be considered a verb. Francis and Yuasa (2008: 74)
cite ni kurabe-te as an example to support the idea that “semantic properties can change
faster than syntactic properties in the process of grammaticalization.”
18
Now, let us consider our main theme, the pragmatic aspects of implicit comparison. In the
following sections, we will consider the status of the implicatures in implicit comparison and
how they arise. Implicit comparison has implicatures that explicit comparison does not have.
Each of the sentences in (29) contains two implicatures in (30):
(29) a. Compared to Tom, Jim is tall.
b. Tom-ni kurabe-tara Jim-wa se-ga takai.
Tom-DAT compare-COND Jim-TOP height-NOM tall
„Compared to Tom, Jim is tall.‟
(30) a. Tom is not tall.
b. Jim is not definitely tall. (possibly borderline)
Figures 4 and Figure 5 diagram the situations that the sentences in (29) denote:
Figure 4: Situation 1 Figure 5 : Situation 2
Tallness Tallness
s s Jim border line
Jim
Tom Tom
[s = contextually determined standard]
In order to investigate the source of the two implicatures in (29), we will propose two
potential explanations of these implicatures: the symmetrical approach and the
asymmetrical approach. In the symmetrical approach, the two implicatures in implicit
comparison are considered to arise from a single processing principle, the economy of
standard. In this view, both of the implicatures have the same status because they are
derived simultaneously.
In the asymmetrical approach, on the other hand, the two implicatures in implicit
comparison are considered to have a different status from each other. This approach is
asymmetrical because in it, the derivation of the implicature from the main clause is
considered to „depend on‟ the existence of the implicature from the standard. We will argue
for the former approach.
19
Before considering the analyses involved in each approach in detail, let us carefully
examine the empirical facts concerning the two implicatures in implicit comparison.
4.1 Implicature from the standard
As mentioned in the introduction, there is an implicature that the NP of the comparative (or
conditional) clause is construed as low on the scale of the gradable predicate (e.g. tallness):
(31) Compared to Tom, Jim is tall.
→Tom is not tall.
The sentence becomes unnatural if we use a standard that is construed as high:
(32) Compared to a gymnast , Tom is tall.
??a basketball player
(33) Compared to a homeless person , Jim is rich.
??a company executive
(34) Compared to Alaska , it is warm here.
??Florida
The reason why one sentence in each of the pairs in (32) through (34) is odd is that there is a
conflict between its low scale implicature and our encyclopedic knowledge.
Notice that if the main sentence in the construction is negative, the scale is reversed:
(35) Compared to a basketball player Sam is not tall.
??a gymnast short
In (35), a basketball player is construed as „low‟ on the scale of „not tallness‟ or „shortness.‟
That is to say, the scale is reversed from tallness to shortness if a sentence is negative or if a
positive sentence contains a „negative adjective‟ like short, which is antonymous to tall, as
shown in Figure 6:14
14 „Negative adjectives‟ are different from „positive adjectives‟ in two ways. First, negative
adjectives license negative polarity items, but positive adjectives do not, as in (i):
(i) It was {foolish/*clever} of her to even bother to lift a finger to help.
20
Figure 6: A scale of shortness/not tallness
Shortness/not tallness
s
Sam
Basketball player
A basketball player is construed as low on this scale. Therefore, (35) is not a counterexample
of the low scale implicature of the compared to clause.
4.2 Implicature from the main clause
Let us now consider the other implicature, i.e. the implicature from the main clause:
(36) Compared to Tom, Jim is happy.
→ Jim is not definitely happy.
This implicature shows that compared to, unlike the morphological comparative, implies
that the (unmodified) positive form is false or borderline for the subject.15 This is further
Second, negative adjectives permit downward entailment while positive adjectives permit
upward entailment, as shown in (ii) and (iii):
(ii) It„s dangerous to climb Mt. Everest. It‟s dangerous to climb Mt. Everest in winter.
(iii) It‟s safe to climb Mt. Hakone. It‟s safe to climb Mt. Hakone in winter.
The two properties are closely related to each other (see Seuren, 1978; Ladusaw, 1979;
Linebarger, 1980; Kennedy, 2001 for detailed discussion of polar adjectives). 15 There is a similarity between the implicature from the main clause in implicit comparison
and the invited inference in if...then conditionals (Geis and Zwicky, 1971) in terms of
negativity. (ia) has a negative implicature like (ib):
(i) a. If you mow the lawn, I‟ll give you $5.
b. → If you don‟t mow the lawn, I won‟t give you $5.
Conditional perfection―the tendency to move from (ia) to (iib)―is taken to be an instance of
invited inference. There is a debate over where the invited inference comes from. There are
two main approaches: the R/I-based approach (Atlas and Levinson, 1981; Horn, 1989, 2000,
among others) and the Q-based approach (Van der Auwera, 1997a, b; Schwenter, 1999,
among others). See also Dancygier and Sweetser (1997) for alternative account of the
21
illustrated by the fact that in implicit comparison one cannot signal the possibility of going
higher up the scale by using the „suspender‟ if not (Horn, 1972), but it is possible in explicit
comparison. Compare examples (37a) and (37b):
(37) a. *Compared to Tom, Jim is happy, if not (more) ecstatic.
b. Jim is happier than Tom, if not more ecstatic.
(cf. Jim is happy if not ecstatic.)
(37a) is not acceptable because there is a negative implicature that „Jim is not definitely
happy‟ in implicit comparison. Not happy and ecstatic are not on the same scale (a Horn
scale), because they are not equally lexicalized (Horn, 1972; Grice, 1975; Gazdar, 1979;
Levinson, 1983, 2000).
However, if we add even to the compared to clause, the sentence becomes natural:16
(38) Even compared to Tom, Jim is happy, if not (more) ecstatic.
This is because even cancels the preexisting negative assumption that “Jim is not happy”
and makes the truth-value of the main clause true. In (38), happy and ecstatic are ordered
along the same Horn scale; therefore, the sentence is natural.
4.3 Borderline case
The implicatures in the main clause of implicit comparison are not uniform. That is to say,
the truth-value of the main clause does not always have to be clearly false. It can be
borderline:
(39) (Context: Mary witnesses a male committing a crime in the street. A police officer
asks her whether the suspect is tall, but she cannot decide whether or not he is.)
Mary: Well, compared to my brother, the suspect is tall.
(a) →The suspect is not tall.
(b) → It is possible that the suspect is not tall.
phenomenon within the mental-spaces framework. Since this paper focuses on the
pragmatics of comparison, we will not get into the debate over the inferences of if...then
conditionals. Notice that a contextually determined standard is not required in the if...then
conditionals. 16 I would like to thank a member of the audience at the 2007 LSA Annual Meeting for
pointing out this fact to me.
22
In this context, the speaker seems to imply (39b) rather than (39a). The implicature of (39b)
is epistemically weaker than (39a). This example shows that the implicatures in the main
clause of an implicit comparative are not uniform.
4.4 Scalar conversion
There is a phenomenon of scalar conversion from low scalar construal to high scalar
construal. This section argues that even and the free choice any cause the phenomenon.
4.4.1 Even and -temo
As we have seen in the previous section, if the scalar additive particle even is inserted into a
clause with compared to, the standard of comparison is construed as high rather than low
on a given scale, as shown in (40):
(40) Even compared to {a basketball player/?? a gymnast}, Jim is tall.
We can state the presupposition of (40) with a basketball player as a standard of comparison
as follows17:
(41) Presupposition of even compared to a basketball player, Jim is tall.
(a) x[C(x) x ≠ a basketball player Jim is tall, compared to x]
(Existential presupposition)
(b) x[x ≠ a basketball player → unlikelihood (Jim is tall compared to a basketball
player) > unlikelihood (Jim is tall compared to x)]
(Scalar presupposition)
(41a) says that there is some x in the given context C and x is not a basketball player, and
Jim is tall, compared to x. (41b) says that the proposition that „Jim is tall compared to a
basketball player‟ is more unlikely than the proposition that „Jim is tall compared to x‟.
In Japanese, the concessive conditional marker -temo converts the scalar construal
from low to high:
17 If (40) is a negative sentence, there will be two possible explanations for the
presupposition of the sentence, one deriving from scope theory (Karttunen and Peters, 1979;
Wilkinson, 1996; Nakanishi, 2006; among others) and the other deriving from polarity
theory (Rooth, 1985; Rullmann, 1997; Giannakidou, 2007; Yoshimura, 2007, among others).
Roughly speaking, scope theory posits only one lexical item for even; its scope interaction
with other operators (e.g. negation) accounts for the positive and negative cases of even.
Polarity theory, on the other hand, posits at least two distinct lexical items for even.
23
(42) {Basuketto/ ?? taisoo} sensyu-ni kurabe-temo Jim-wa se-
Basketball/gymnastics player-DAT compare-CONCE.COND Jim-TOP height-
ga takai.
NOM tall
„Even compared to {a basketball player/ ?? a gymnast}, Jim is tall.‟
(42) is natural with a basketball player, but not with a gymnast, as the standard of
comparison.18
The scale of (42) with a basketball player as the standard of comparison is diagramed in
Figure 7:
Figure 7: Scale of (42) with a basketball player
Tallness
Jim
[a basketball player]F
s
Here, the concessive conditional sentence with a basketball player as a standard of
comparison cancels the preexisting assumption that „Jim is not tall‟ and raises Jim‟s level
above that of a basketball player.
4.4.2 Free-choice any and wh-morpheme plus –temo in Japanese
The free-choice any also overrides low scale implications. (43a) is semantically similar to
(43b)19:
18 Notice that in Japanese the scalar additive particle sae cannot combine with the
conditional marker tara „if ‟:
(i) *{Basuketto/ taisoo} sensyu-ni kurabe-tara-sae Jim-wa se- ga takai.
Basketball/gymnastics player-DAT compare-COND-even Jim-TOP height-Nom tall
„Even compared to {a basketball player/ ?? a gymnast}, Jim is tall.‟
In English it is possible to represent a concessive conditional meaning by combining the
conditional marker if with even (i.e. even if). But in Japanese we must use the marker temo
in order to express a concessive conditional meaning (See Fujii, 1989). 19 I would like to thank Chris Kennedy for pointing out this scalar conversion phenomenon.
24
(43) a. Compared to anyone, Jim is tall.
b. Even compared to the tallest person, Jim is tall.
Lee and Horn (1995) argue that a free-choice any is semantically equivalent to a „generic
indefinite‟ (Diesing 1992) and it can be paraphrased by even plus a (contextually
appropriate) superlative. Thus, (43a) can be paraphrased by (43b). This any is a free-choice
any because it can be modified by almost or absolutely:20
(44) Compared to almost anyone, Tom is tall.
Notice that the meaning of free choice is expressed by using a wh-morpheme and the
concessive marker -temo „even if‟ in Japanese:
(45) Dare-to kurabe-temo Tom-wa se -ga takai.
Who-with compare-CONCE.COND Tom-Top height- Nom tall
„Compared to anyone, Tom is tall.‟
5 Symmetrical approach to the implicatures of implicit comparison
Section 4 has pointed out various pragmatic aspects of implicit comparison. This section will
analyze the two implicatures of implicit comparison more theoretically. We will consider the
status of these two implicatures and how they are derived. We will posit two approaches to
solve these problems, the symmetrical (economy-based) approach and the asymmetrical
(dependency) approach.
5.1 Implicit comparison vs. positive form
Before introducing the symmetrical (economy-based) approach to implicit comparison, we
need to consider the following question: Why does the speaker use implicit comparison,
rather than a simple sentence with a positive form like (46b) or (46c) in a given context?:
20 As Lee and Horn (1995: 4) argue, if any is interpreted universally, it can be modified by
absolutely but if it is interpreted existentially, this is not possible:
(i) a. Alfred will eat any food. (Fauconnier, 1975)
b. Alfred will eat absolutely/almost any food. (Fauconnier, 1975)
(ii) a. Alfred didn‟t eat any food.
b. *Alfred didn‟t eat absolutely/almost any food.
25
(46) a. Compared to Tom, Jim is tall.
b. Jim is not tall.
c. Jim is tall.
Obviously, (46a) is longer and more complex than (46b) and (46c). We can explain the
motivation behind implicit comparison by using Horn‟s Division of Pragmatic Labor. Horn
(1989: 304) defines this concept as follows:
(47)
The use of a longer, marked expression in lieu of a shorter expression involving
less effort on the part of the speaker signals that the speaker was not in a position
to employ the simpler version felicitously. (...) There is a correlation between the
stylistic naturalness of a given form, its relative brevity and simplicity, and its use
in stereotypic situations; this reflects the operation of the R Principle. The
corresponding periphrastic forms, stylistically less natural, longer, and more
complex, are restricted, via Q-based implicature, to those situations outside the
stereotype, for which the unmarked expression could not have been used
appropriately. (Horn, 1989: 304)
In (46) we can say that the speaker of (46a) uses implicit comparison because he or she
thinks that shorter and unmarked expressions like (46b) and (46c) cannot express indirect
or weak negative implicature. In other words, the speaker of (46a) is trying to change the
evaluation „indirectly‟ by introducing a new standard of comparison.21
5.2 Economy principle
Horn‟s Division of Pragmatic Labor can explain why the speaker chose implicit comparison
rather than the positive form of the adjective. But this does not explain the mechanism by
which the implicature of implicit comparison arises.
21 The classical example of the Division of Pragmatic Labor is the phrasal causative vs. the
lexical causative. McCawley (1978) points out that sentences (i) and (ii) are different in
meaning:
(i) He caused the sheriff to die.
(ii) He killed the sheriff. (McCawley, 1978: 249)
McCawley (1978: 249) argues that (i) “would be an inappropriate thing to say if the person
in question shot the sheriff to death, since there is an alternative available involving a less
complex surface structure.” He argues that a periphrastic causative as in (ii) is interpreted
as referring to „indirect‟ causation.
26
In the symmetrical approach, the economy principle can explain the nature of
implicatures in implicit comparison. This principle ensures that a truth-conditional
interpretation is achieved with „no more effort than is necessary‟ as shown in:
(48) Economy of „standard of comparison‟: It is a violation of economy to use a special
form to introduce a new standard, if the truth-value of the main proposition in
implicit comparison does not change.
This principle says that it is preferable to have as few standards as possible. That is, if we
introduce a new standard, the truth-value of the main proposition has to change. Therefore,
if we bother to introduce a new explicit standard, there must be good reason for doing so.
Therefore, it is possible to say that the economy principle is a natural extension of Horn‟s
Division of Pragmatic Labor. I will argue that this economy principle is different from
(Neo)-Gricean maxims in that it is a truth value oriented principle.
Recall that the situations in which it is acceptable to employ the implicit comparison
„compared to x, y is tall‟ are those diagrammed in Figure 8 and Figure 9:
Figure 8: Situation 1 Figure 9: Situation 2
s s y borderline zone
y
x (new s) x (new s)
The economy principle can be tested by using the following scheme:
(49) Schema: a. (Compared to x, [y is Gradable P]). (English)
b. (x-ni kurabe-tara [y-wa Gradable P]) (Japanese)
Within the domain of the square brackets, i.e. […], the truth value of proposition in the
main clause is calculated without including the comparative clause. In this situation, the
truth value is calculated based on a contextually determined standard. (For example, the
truth value of the proposition „this building is tall‟ is calculated based on a contextually
determined standard.) Within the domain indicated by the parentheses, i.e. (…[…]), the
truth value of the proposition in the main clause is calculated based on the explicit standard
27
of comparison, namely x. Table 1 shows the possible patterns of truth value in implicit
comparison:
Table 1: The possible patterns of truth value in implicit comparison
Truth value of […] Truth value of (…[…]) Result
a. F T Good
b. ? T Good (borderline case)
c. T F Bad
d. ? F Bad
e. F F Bad
f. T T Bad
Let us first look at the bad cases. Rows (c), (d), and (e) in Table 1 are all bad because the
truth value of the entire sentence (…[…]) is false and violates the maxim of quality: „Do not
say what is considered to be false.‟ For example, (50) is an example of row (c):
(50) (Context: Tom is tall but he is not taller than a basketball player.)
Compared to a basketball player, Tom is tall.
(51) is an example of row (d):
(51) (Suppose that a soccer player is a borderline case.)
Compared to a basketball player, a soccer player is tall.
(52) is an example of row (e):
(52) Compared to a basketball player, a gymnast is tall.
Row (f) is important. It is bad not because it violates the maxim of quality, but because it
violates the general principle of economy of standard. (53) is an example of row (f):
(53) (Context: Tom is the tallest man in the United States)
?? Compared to a basketball player, Tom is tall.
One may presume that (53) implies that Tom is extremely tall, but it does not induce such
an implicature. In fact, the sentence is very odd. The reason why it is odd is that the
truth-value of the proposition in the main clause does not change even if we introduce a new
standard, in this case, a basketball player. This sentence violates the principle of economy of
28
standard in (48).22 Another example for row (f) is shown in:
(54) (Context: Tom is a tall basketball player.)
?? Compared to a gymnast, Tom is tall.
(54) would be odd in this context. The speaker does not need to posit the explicit standard,
i.e. a gymnast in this context, because it does not affect the truth value of the proposition in
the main clause. The above discussion suggests that for implicit comparison, all that is
relevant is the truth-value of the proposition of the main clause.
The good rows are (a) and (b). Row (a) is a case where the truth value of the proposition
without an explicit standard is clearly false, but the truth value of the proposition with an
explicit standard is true. Row (b) is a borderline case where the truth value of the
proposition without an explicit standard is false but the truth value of the proposition with
an explicit standard is a borderline.
The economy of standard constraint enables us to explain the nature of the two
implicatures. The principle says that it is preferable to have as few standards as possible.
And if we bother to introduce a new explicit standard, there must be good reason for doing
so. The only situations in which there is good reason to posit a new explicit standard is those
in rows (a) and row (b). Here, the speaker makes a false or borderline proposition true. In
order to do so, we must introduce a new standard at the extreme low end of the given scale.
We cannot introduce a new standard that is not very different from the contextually given
standard, because that would not allow us to change the truth value of the proposition in
the main clause. Because the two implicatures can be explained by a single principle, they
are symmetrical, not asymmetrical.
One potential problem for the economy-based explanation is that it wrongly predicts
that the following examples are not acceptable:
(55) Even compared to a basketball player, Tom is tall.
(56) Compared to anyone, Tom is tall.
22 The oddness in (53) can also be captured in terms of discourse structure (e.g. Roberts,
1998):
(i) a. Is Tom tall?
b. ?? Yes, compared to a basketball player, Tom is tall.
(ib) sounds odd (and redundant) because the speaker in (ib) is answering to the question
under discussion by introducing the unnecessary standard (i.e. a basketball player). I thank
Yusuke Kubota for bringing this to my attention.
29
In (55) and (56), the truth value of the main clause and that of the entire sentence are both
true. Therefore, the principle of economy of standard wrongly predicts that they are
ill-formed.
Why are sentences (55) and (56) acceptable? Obviously because of the presence of even
and any. Recall that if even or any are used in implicit comparison, the sentences are
interpreted as superlatives that give rise to a universal meaning (Section 4.4.2). Thus, the
sentences in (55) and (56) are equivalent to saying that Tom is the tallest man (in the given
context). Clearly, superlatives are different from the positive sentences (Tom is tall) in terms
of quantification. I would argue that the economy of standard does not apply to (55) and (56),
because implicit comparison with even/any is not a pure „adjectival‟ sentence in terms of
quantification. (55) and (56) do more than just posit a new standard with respect to Tom‟s
height.
5.3 Economy of standard of comparison: Whose economy?
The economy principle is neither stored in the particular lexicon nor is it one of the (neo-)
Gricean maxims. It is well-known that the definition of the R-principle and the Q principle
are based on economy.
(57) a. The R-principle: Say no more than you must. (speaker‟s economy)
b. The Q-principle: Say as much as you can. (hearer‟s economy)
R-implicature is an UPPER bounding correlate of the law of least effort, which dictates
minimization of form. (58) and (59) are examples of R-based implicauture:
(58) Tom broke a finger yesterday.
→ The finger was one of Tom‟s.
(59) Tom ate the cake.
→Tom ate the whole cake. (Harnish, 1976)
In these examples, „p implies more than p.‟ Therefore, R-implicature is basically „positive‟ in
nature.
Although the economy of standard principle is similar to the R-principle in the sense
that they both use the law of least effort, the notion of economy in this principle seems to be
neither a speaker-oriented economy nor a hearer-oriented economy. The principle of the
economy of standard is a truth-value oriented economy:
30
(60) Economy Principle: Calculate the truth-value with no more effort than is
necessary.
The principle of economy of standard plays an important role in our communication. Based
on the above discussion, I would like to propose that there is a third type of implicature, the
computational implicature, as shown in figure 10:
Figure 10: Three types of implicature
Conversational implicature
Implicature Computational implicature
Conventional implicature
The computational implicature is the product of the application of the principle of economy
of standard.
6 Asymmetrical approach to the pragmatics of implicit comparison
This section focuses on the alternative asymmetrical approach. In this approach, an
asymmetrical strategy is used for the two implicatures because the implicature from the
main clause depends on the implicature from the standard. If the implicature from the
standard is not satisfied, the implicature from the main clause does not arise.
6.1 Is the implicature from the standard conventional?
Sentence (61) implies that Tom is not tall:
(61) Compared to Tom, Jim is tall.
→ Tom is not tall.
Where does the low inference implicature come from? Does it derive from the lexicon
(conventionalized implicature) or from one of the cooperative principles (conversational
implicature)? It is possible to check whether this implicature is conventionalized or not by
using the tests of cancelability and detachability. If an implicature is detachable, but not
cancelable, the implicature is conventional, but if it is both nondetachable and cancelable, it
31
is conversational (Grice, 1989; Sadock, 1978).
(62) Cancelability:
Compared to Tom, Jim is tall. #And since Tom is tall, that makes Jim really tall.
The fact that the low construal implicature is difficult to cancel suggests that the
implicature is conventional. Notice that the cancellation of the implicature from the
standard automatically triggers the cancellation of the implicature from the main clause.
The reason why (62) is not natural may be that the sentence tries to cancel the two
implicatures simultaneously.
Next, let us test the implicature by using the detachability test:
(63) Detachability:
a. Compared to Tom, Bill is tall. (→Tom is short.)
b. If Bill is compared to Tom, he is tall. (→Tom is short)
c. Considering Tom, Bill is tall. (→Tom is short.)
Examples (63a-c) all imply that „Tom is short.‟ This might lead us to conclude that the
implicature is not detachable and is therefore a conversational implicature. Therefore, there
seems to be a paradox:
( 64 ) The cancelability test suggests that the inference is conventional, but the
detachability test suggests that the inference is conversational.
It seems that there are two possible approaches to this paradox. The first approach is to
consider that the implicature from the standard is conversational because the implicature
from the standard is more cancelable than the typical conventional implicature:
(65) a. Tom is a politician, therefore corrupt. ## There is no connection between
politicians and corruption, however.
b. Compared to Tom, Jim is tall. # And since Tom is tall, that makes Jim really
tall.)
(65b) sounds more natural than (65a), although it is not perfectly natural. If we take the
view that the implicature from the standard is conversational, we need to account for its
source by using one of Grice‟s Maxims.
32
The second approach is to consider that the implicature from the standard is
conventional. This approach emphasizes that the detachability test is not always
unproblematic. To test for non-detachability, we need to posit a set of synonymous
expressions, which should share the same implicature. However, as Sadock (1978) points
out, the implicature can actually be part of the semantic content of each member of that set.
Therefore, it is, in principle, possible to argue that the constructions compared to,
considering, and if…then all possess the same kind of „conventional implicature.‟ These
three constructions all share a conditional flavor.
Notice, however, that we cannot easily say that an implicature is part of the phrase
compared to. Observe the following examples:
(66) a. Compared to Tom, Bill is tall.
b. Compared to Tom, Bill is taller.
Although sentence (66a) implies that „Tom is not tall‟, sentence (66b), to the extent that it is
acceptable, does not imply such a negative implication (Chris Kennedy p.c.). This suggests
that it is not the simple expression compared to but the construction „compared to x, y is a
gradable predicate‟ that possesses the conventional implicature.
The above discussion suggests that it is not clear whether the implicature from the
standard is conventional or conversational. (We will come back to this problem later. I will
argue that this is not a problem for the economy-based account because according to that
theory, the implicature of implicit comparison is neither conversational nor conventional
but computational.)
6.2 Implicature from the main clause is conversational
Now let us consider the implicature from the main clause. Sentence (67a) implies (67b):
(67) a. Compared to Tom, Jim is tall.
b. → Jim is not definitely tall. (possibly borderline)
This implicature shows that compared to, unlike the morphological comparative, implies
that the (unmodified) positive form is false (or unknown) of the subject. This implicature
seems to be conversational, because it is both cancelable and non-detachable:
(68) Cancelability test (context: the speaker is asked weather Jim is tall.)
Compared to Tom, Jim is tall. In fact, Jim is tall compared to anyone.
33
(69) Non-detachability test
a. Compared to Tom, Jim is tall.
b. Considering Tom, Jim is tall.
c. If Jim is compared to Tom, Jim is tall.
In (68) the speaker reevaluated Jim‟s tallness and cancelled the negative implicature
induced by the previous utterance. As for the non-detachability test, any other way of
expressing the literal content of (69a) would license the same implicature that „Jim is not
definitely tall.‟ Because the implicature is both cancelable and non-detachable, it is safe to
consider this implicature to be conversational.23 Then how can the implicature arise? It
seems that the implicature derives from Maxim of Quantity: Say as much as you can. We
can say that the Q-implicature arises based on a „substitutive‟ scale on the assumption that
compared to is a kind of focus phrase24:
(70) The scale for compared to Tom, Jim is tall.
Compared to [an ordinary standard], Jim is tall.
Compared to [Tom]F, Jim is tall.
By saying that compared to Tom, Jim is tall, the speaker implies that „compared to an
ordinary standard, Jim is NOT tall‟, via the Q-principle.25 Compared to is a kind of a focus
construction because it induces an alternative element x to a focused element (Rooth, 1985,
1992). However, it is important to notice that we must posit an additional constraint: that x
is a contextually determined standard or a person above the standard. Therefore, we can
say that the sentence compared to [Tom]F, Jim is tall presupposes (71):
(71) (a) x[C(x) x ≠ Tom]
(b) x ≥ contextually determined standard
C in (71a) is a contextual domain variable that is fixed by context.
23 The implicature in the main clause is also reinforceable, which provides further support
for the argument that the implicature is conversational:
(i) Compared to Tom, Bill is tall but Bill is not definitely tall.
24 The fact that x in „compared to x‟ cannot be a de-accented noun also supports the idea that
it is a kind of focus construction. 25 The alternative element may be a person who corresponds to a contextual ordinary
standard.
34
6.3 The dependency of the implicature from the main clause
Notice that the Q-implicature does not arise without the implicature from the standard. If
the focused element is higher on the scale than its alternative, the Q-implicature does not
arise. That is, the conversational implicature depends on the conventional implicature. For
example, the following scale is bad because there is no higher element.
(72) Bad scale: (No Q-implicature)
The scale for ??Compared to a basketball player, Jim is tall.
Compared to F [a basketball player], Jim is tall.
Compared to [an ordinary standard], Jim is tall.
Since the sentence compared to a basketball player, Jim is tall does not satisfy the
implicature from the standard, the implicature from the main clause cannot arise.
7 Which approach is preferable, the asymmetrical approach or the symmetrical approach?
We have considered two possible explanations for the implicatures in implicit comparison, a
symmetrical approach and an asymmetrical approach. We can summarize each approach as
follows:
(73) Asymmetrical approach:
(a) Low scale implicature from the standard can be conventional because it is not
cancelable (But the result changes to conversational, if we invoke the
detachablility test.)
(b) The implicature from the main clause is conversational (a quantity
implicature).
(c) The implicature in (b) depends on the implicature in (a).
(74) Symmetrical approach (economy-based approach): The two kinds of implicature
are explained simultaneously by the economy principle, the economy of standard of
comparison.
Which approach is preferable? I will argue that the symmetrical approach is preferable for
the following reasons.
First, the asymmetrical approach cannot explain the nature of the implicature from the
comparative clause clearly. A paradox arises, because it is conventional from the standpoint
35
of the cancelability test, but it is conversational from the standpoint of the detachability
test.
Second, the symmetrical approach (economy-based approach) can explain both kinds of
implicatures in implicit comparison simultaneously. It does not have to posit a step-by step
derivation between the two implicatures.
The third advantage is concerned with the implicature from the main clause. The
asymmetrical approach views this implicature as a scalar implicature. This approach has to
stipulate that the alternative element (other than the NP) in the comparative clause must
be a contextually determined standard in order to induce a proper scalar implicature. Let us
consider the situation diagrammed in Figure 11:
Figure 11
Tallness
s
Jim (target)
Bob
Tom (standard)
In the asymmetrical approach, an alternative element cannot be an element that is between
the standard of comparison and the target of the comparison on a given scale. This is
because if we posit a situation like the one in Figure 11, the asymmetrical approach wrongly
induces the Quantity implicature: that „compared to Bob, Jim is not tall.‟ This implicature is
in conflict with the situation in Figure 11, at least semantically.26
However, the symmetrical approach is not problematic for the situation in Figure 11,
because this approach focuses on the truth-value of the main clause, so it does not matter
whether there is a person between Tom and Jim on the given scale of height.
A fourth advantage for the economy-based approach is its generality. This approach can
also account for the for PP phrases. The English for PP phrase also works like implicit
comparison: 27
26 Notice that the sentence Compared to Bob, Jim is not tall‟ can be natural in the situation
in Figure 9, if the speaker thinks that there is not a great difference in height between Bob
and Jim. If there is in fact only a tiny difference between Bob and Jim, it is difficult to
partition Jim into the positive extension and Bob into the negative extension. This can be
another example of Crisp Judgment. 27 Strictly speaking, (75) is not a „regular‟ comparative. As Kennedy (2007a) argues, the
36
(75) Jim is tall for a gymnast.
→ Jim is not definitely tall. (possibly a border line)
→ A gymnast is not tall.
(75) is similar to implicit comparison in the sense that (75) implies negative implicatures.
The for PP phrase is just like compared to. It introduces a new standard of comparison. The
principle of economy of standard enables us to explain why there is a difference in
acceptability in (76) without assuming the conventional meaning:
(76) Jim is tall for {a gymnast/?? a basketball player}.
According to the principle, it is a violation of economy to use the for PP phrase if the new
explicit standard does not change the truth value of „Jim is tall‟. Thus, the only situation in
which it is possible to use the for PP phrase is one in which situation in which the truth
value of the proposition „Jim is tall‟ (which is calculated based on a contextual standard) is
considered to be a false or borderline, and the truth value of the proposition „Jim is tall‟
(which is calculated based on for PP) is considered to be true. That is why the NP in the for
PP phrase has to have a low scale value, rather than a high scale value.
Based on the above arguments, I conclude that the economy-based approach is
preferable. The theoretical implication is that there is a third type of implicature: a
computational implicature.
8 Conclusion
This paper has developed a pragmatic analysis of implicit comparison. There are two
implicatures in implicit comparison. One is the implicature that the standard of comparison
is construed as low on a given scale, and the other is that the proposition in the main clause
is false or unknown. We have considered the two possible approaches to these implicatures,
the symmetrical approach and the asymmetrical approach. From the standpoint of the
asymmetrical approach, the implicature from the main clause depends on the implicature
from the standard. From the standpoint of the symmetrical approach, on the other hand,
the two implicatures are both seen as deriving from a single principle, the principle of
economy of standard. This principle states that it is a violation of economy to introduce a
new standard if the truth value of the main proposition in implicit comparison does not
sentence x is A for an NP presupposes that x is an NP.
37
change. That is to say, in implicit comparison, the speaker converts the proposition in the
main clause, which is construed as false or unknown, into a proposition that is construed as
true by introducing a new standard that is construed as extremely low on a given scale.
Notice that the principle of economy of standard enables us to explain the nature of both
implicatures at the same time, so that it is unnecessary to consider that the two
implicatures have a different pragmatic status. I conclude from this that the symmetrical
approach is preferable.
The reason why the principle of economy of standard does not apply to explicit
comparison is that in that case, the truth value of the proposition in the main clause does
not matter. Japanese is interesting in this regard, because the fact that it lacks comparative
morphology makes the main clause in implicit and explicit comparatives syntactically
identical. However, although no comparative morphology is used in the environment „x is
A-er than y.‟ in Japanese, the language does have a distinction between explicit and implicit
comparisons.
The notion of economy has so far played an important role in pragmatics. It is used to
explain the difference of the Q-principle and the R-principle. The Q-principle is considered
to be a hearer-based economy and the R-principle is considered to be a speaker-based
economy (Horn, 1989). However, the economy principle discussed in this paper is not a
speaker/hearer-based economy. Instead, it is concerned with the economy of a
truth-conditional interpretation. This paper has proposed the existence of a third type of
implicature, a computational implicature that derives neither from a lexicon nor from a
co-operative principle. I hope this paper will shed a new light on the nature of implicature in
natural language.
Acknowledgements
I am very grateful to Chris Kennedy for valuable discussions and insightful comments
about this material. Thanks are also due to Anastasia Giannakidou, Jerry Sadock, and
Yusuke Kubota for their detailed comments and suggestions on the earlier version of this
paper. Finally, I would like to express my gratitude to Peter Alrenga, Jay Atlas, Kent Bach,
Gennaro Chierchia, Malcolm Elliott, Michael Yoshitaka Erlewine, Tommy Grano, Chris
Potts, Harumi Sawada, Jun Sawada, Keiko Yoshimura, the members of Semantics Lab at
the University of Chicago, and the reviewers for Journal of Pragmatics for their valuable
comments and suggestions. Parts of this paper were presented at the 9th Annual meeting of
the Pragmatics Society of Japan, 2007 LSA Meeting and the 1st Midwest Workshop on
38
Semantics at the University of Chicago in 2006 and I thank the audiences for their very
helpful feedback. Any remaining errors are of course my own.
References
Atlas, Jay D., Levinson, Stephen C., 1981. It-cleft, informativeness and logical form. In:
Cole, P. (Ed.), Radical Pragmatics. Academic Press, New York, pp. 1-61.
Beck, Sigrid, Oda, Toshiko, Sugisaki, Koji, 2004. Parametric variation in the semantics
of comparison: Japanese vs. English. Journal of East Asian Linguistics 13, 289-344.
Dancygier, Barbara, Sweetser, Eve, 1997. Then in conditional constructions. Cognitive
Linguistics 8, 109–136.
Diesing, Molly, 1992. Indefinites. MIT Press, Cambridge, Mass.
Fauconnier, Gilles, 1975a. Polarity and the scale principle. In: Grossman, R.E., San L. J.,
Vance, T.J. (Eds.), Papers from the Eleventh Regional Meeting of the Chicago
Linguistic Society, Chicago Linguistic Society, Chicago, pp. 188-199.
Fauconnier, Gilles, 1975b. Pragmatic scales and logical structure. Linguistic Inquiry 6,
353-375.
Francis, Elaine J., Yuasa, Etsuyo, 2008. A multi-modular approach to gradual change in
Grammaticalization. Journal of Linguistics 44, 45-86.
Fujii, Seiko, 1989. Concessive conditionals in Japanese: a pragmatic analysis of the
S1-TEMO -S2 construction. In: Kira, H, Meacham M, Shapiro, R. (Eds.),
Proceedings of the 15th Annual Meeting of the Berkeley Linguistics Society.
Berkeley Linguistics Society, Berkeley, pp. 291-302.
Gazdar, Gerald, 1979. Pragmatics: Implicature, Presupposition and Logical Form.
Academic Press, London.
Geis, Michael L., Zwicky Arnold M., 1971. On invited inferences. Linguistic Inquiry 4,
561-566.
Giannakidou, Anastasia, 2007. The landscape of EVEN. Natural Language and Linguistic
Theory 25 (1), 39-81.
Grice, Paul H., 1975. Logic and conversation. In: Cole, P., Morgan, J. (Eds.), Syntax and
Semantics Volume 3, Speech Acts. Academic Press, New York, pp. 41-58.
Grice, Paul H., 1989. Studies in the Way of Words. Harvard University Press, Cambridge,
MA.
Harnish, Robert M., 1976. Logical form and implicature. In: Bever T. G., Katz J. J.,
Langendoen T. D. (Eds.), An Integrated Theory of Linguistic Ability. New York
39
Crowell, New York, pp. 313-319.
Hirschberg, Julia, 1985. A Theory of Scalar Implicature. Ph.D. dissertation, University of
Pennsylvania.
Horn, Laurence R., 1972. On the Semantic Properties of Logical Operators in English.
Indiana University Linguistics Club, Bloomington, IN.
Horn, Laurence R., 1984. Toward a new taxonomy for pragmatic inference: Q-based and
R-based implicature. In: Schiffrin D. (Ed.), Meaning, Form, and Use in Context:
Linguistic Applications. Georgetown University Press, Washington, pp. 11-42.
Horn, Laurence R., 1989. A Natural History of Negation. University of Chicago Press,
Chicago, IL.
Horn, Laurence R., 2000. From if to iff: conditional perfection as pragmatic strengthening.
Journal of Pragmatics 32, 289-326.
Karttunen, Lauri, Peters, Stanley, 1979. Conventional implicature. In: Oh C.K. and Dinnen
D A. (Eds.), Syntax and Semantics. Vol 11 Presupposition. Academic Press, New
York, pp. 1-56.
Kay, Paul, 1990. Even. Linguistics and Philosophy 13, 59-111.
Kennedy, Christopher, 2007a. Vagueness and grammar: the semantics of relative and
absolute gradable adjectives. Linguistics and Philosophy 30(1), 1-45.
Kennedy, Christopher, 2007b. Standards of comparison. Handout of Colloque de Syntaxe et
Sémantique à Paris.
Kennedy, Christopher, in press. Modes of comparison. In: Elliott, M., Kirby, J., Sawada, O.,
Staraki, E., Yoon, S. (Eds.), Paper from the 43rd Annual Meeting of the Chicago
Linguistic Society. Chicago Linguistic Society, Chicago.
Kennedy, Christopher, 2001. On the monotonicity of polar adjectives. In: Hoeksema, J.,
Rullmann, H., Sánchez-Valencia, V., van der Wouden, T. (Eds.), Perspectives on
Negation and Polarity Items. John Benjamins Publishing,
Amsterdam/Philadelphia, pp. 201-221.
Kennedy, Christopher, McNally, Louise, 2005. Scalar structure, degree modification, and
the semantics of gradable predicates. Language 81(2), 345-381.
Klein, Ewan, 1980. A semantics for positive and comparative adjectives. Linguistics and
Philosophy 4, 1-45.
Ladsaw, Willian, 1979, Polarity as Inherent Scope Relations. Ph.D. dissertation, University
of Texas, Austin.
Lakoff, George, 1972. Hedges: a study in meaning criteria and the logic of fuzzy concepts. In:
Peranteau, P. N., Levi, N. J., Phares, G. C. (Eds.), Papers from the Eighth Regional
Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chicago, pp.
40
183-228.
Lehmann, Christian, 1988. Towards a typology of clause linkage. In: Haiman, J.,
Thompson, S. A. (Eds.), Clause Combining in Grammar and Discourse. John
Benjamins, Amsterdam, pp. 181-225.
Lee, Young-Suk and Horn, Laurence, 1995. Any as indefinite plus even. Manuscript, Yale
University.
Levinson, Stephen C., 1983. Pragmatics, Cambridge University Press, Cambridge.
Levinson, Stephen C., 2000. Presumptive Meanings: The Theory of Generalized
Conversational Implicature. MIT Press, Cambridge, Mass.
Linebarger, Marcia, 1980. The Grammar of Negative Polarity. Ph.D. dissertation,
Massachusetts Institute of Technology.
Martin, Samuel, 1975. A Reference Grammar of Japanese. Yale University Press, New
Haven and London.
McCawley, James. D., 1978. Conversational implicature and the lexicon. In: Cole, P. (Ed.),
Syntax and Semantics Volume 9: Pragmatics, Academic Press, New York, pp.
245-259.
Muraki, Shinjiro, 1991. Nohongo Dooshi no Syosoo (Aspects of Japanese verbs), Hitsuji
Syoboo, Tokyo.
Nakanishi, Kimiko, 2006. The semantics of even and negative polarity items in Japanese.
In: Baumer D., Montero D., Sxanlon M. (Eds.), Proceedings of the 25th West Coast
Conference on Formal Linguistics. Cascadilla Proceedings Project, Somerville MA,
pp. 288-296.
Roberts, Craige, 1996. Information structure in discourse: toward an integrated formal
theory of pragmatics. In: Jae Hak Yoon & Andreas Kathol (Eds.), OSU Working
Papers in Linguistics, vol. 49: Papers in Semantics, The Ohio State University
Department of Linguistics, Columbus, OH, pp. 91-136. Revised 1998.
Rooth, Mats, 1992. A Theory of focus interpretation. Natural Language Semantics 1, 75-116.
Rooth, Mats, 1985. Association with Focus, Ph.D. dissertation, University of Massachusetts
at Amherst.
Rotstein, Carmen, Winter Yoad, 2004. Total adjectives vs. partial adjectives: scale structure
and higher-order modifiers. Natural Language Semantics 12, 259-288.
Rullmann, Hotze, 1997. Even, polarity, and scope. In: Gibson, M., Wiebe, G., Libben, G.
(Eds.), Papers in Experimental and Theoretical Linguistics 4. University of
Alberta, Edmonton, pp. 40-64.
Sadock, Jerrold M., 1978. On testing for conversational implicature. In: Cole, P. (Ed.),
Syntax and Semantics Volume 9: Pragmatics. Academic Press, New York, pp.
41
281-297.
Sapir, Edward, 1944. Grading: a study in semantics. Philosophy of Science 11, 93-116.
Sawada, Osamu, in press. The comparative morpheme in Modern Japanese. In: The
Proceedings of the 34th Annual Meeting of the Berkeley Linguistics Society.
Berkeley Linguistic Society, Berkeley.
Sawada, Osamu, 2007. Pragmatic aspects of implicit comparison. Paper presented at the
81st Annual Meeting of the Linguistic Society of America. Anaheim, CA.
Sawada, Osamu, 2005. The possible semantic diversity of the comparative constructions in
English and Japanese: a construction-based approach. Paper presented at the 9th
International Cognitive Linguistics Conference. Yonsei University, Korea.
Schwarzschild, Roger, 2005. Measure phrases as modifiers of adjectives, Recherches
Linguistiques de Vincennes 34, 207-228.
Schwenter, Scott A., 1999. Pragmatics of Conditional Marking: Implicature, Scalarity, and
Exclusivity. Garland Publishing, New York.
Seuren, Pieter, 1973. The comparative. In: Kiefer, F., Ruwet, N. (Eds), Generative
Grammar in Europe. Riedel, Dordrecht, pp. 528-564.
Seuren, Pieter, 1978. The structure and selection of positive and negative gradable
adjectives. In: Farkas, D., Jacobson, W., Todrys, K. (Eds.), Papers from the
Parasession on the Lexicon, Chicago Linguistic Society 14. Chicago Linguistic
Society, Chicago, pp. 336-346.
Silverstein, Michael, 1976. Hierarchy of features and ergativity. In: Dixon, R. M. W. (Ed.),
Grammatical Categories in Australian Languages. Australian Institute of
Aboriginal Studies, Canberra, pp. 112-171.
Snyder, William, Wexler, Kenneth, Das, Dolon, 1995. The syntactic representation of degree
and quantity: perspectives from Japanese and child English. In: Aranovich, R.,
Byrne, W., Preuss, S., Senturia, M. (Eds.), The Proceedings of the Thirteenth West
Coast Conference on Formal Linguistics. CSIL, Stanford, pp. 581-596.
Ultan, Russell, 1972. Some features of basic comparative construction. Working Papers on
Language Universals 9, 117-162.
Van der Auwera, Johan, 1997a. Conditional perfection. In: Angeliki A., Dirven, R. (Eds.), On
Conditionals Again. John Benjamins, Amsterdam/Philadelphia, pp. 169-190.
Van der Auwera, Johan, 1997b. Pragmatics in the last quarter century: the case of
conditional perfection. Journal of Pragmatics 27, 261-274.
Wilkinson, Karina, 1996. The scope of even, Natural Language Semantics 4, 193-215.
Yoshimura, Keiko, 2007. Focus and Polarity: Even and Only in Japanese. Ph.D.
Dissertation. University of Chicago.
42
Yuasa, Etsuyo, Sadock, Jerry M., 2002. Pseudo-subordination: a mismatch between syntax
and semantics. Journal of Linguistics 38, 87-111.